Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

Algorithms ID Factors in Parkinson’s Progression

Brain circuits illustration

(HypnoArt, Pixabay)

26 September 2017. Data from two clinical trials analyzed with machine learning algorithms identify factors that predict the decline of motor functions in individuals with Parkinson’s disease. The findings, conducted by a team from the medical informatics company GNS Healthcare in Cambridge, Massachusetts, appear in yesterday’s issue of the journal The Lancet Neurology (paid subscription required).

Parkinson’s disease occurs when the brain produces less of the substance dopamine, a neurotransmitter that sends signals from one neuron or nerve cell to another. As the level of dopamine lowers, people with Parkinson’s disease become less able to control their bodily movements and emotions. Symptoms include tremors, i.e. shaking, slowness and rigidity in movements, loss of facial expression, decreased ability to control blinking and swallowing, and in some cases, depression and anxiety. According to Parkinson’s Disease Foundation, some 60,000 new cases of Parkinson’s disease are diagnosed in the U.S. each year, with more than 10 million people worldwide living with the disease.

GNS Healthcare offers a technology that applies machine learning to health care information, but with a process designed to reveal causal factors in the data, not just associations. As described by the company, that process starts by reverse engineering of data with deep-learning algorithms in large data sets from genomics, electronic health records, demographics, pharmacy and medical claims records, imaging, and mobile devices.

These algorithms yield models with potential causal factors addressing the therapy targets. GNS Healthcare then tests the models with a series of “what-if” simulations to find the optimum solutions, such as best therapies for specific individuals.

The solutions sought in this case by the team led by Jeanne Latourelle, GNS Healthcare’s director of precision medicine, seek to better predict the progression of Parkinson’s disease as it affects motor functions in patients. Not only would the findings better help prescribe more precise treatments for Parkinson’s disease patients, say the authors, but also help design clinical trials of new therapies.

Latourelle and colleagues drew data from two clinical trials of Parkinson’s disease. One study currently underway aims to uncover imaging and biologic indicators of Parkinson’s disease. From that trial’s participants, the GNS Healthcare team sampled data from 312 individuals with Parkinson’s disease and 117 healthy persons for comparison. The researchers then analyzed participants’ genetic, clinical, and demographic data against results on a standard rating scale of motor functions and quality of life for Parkinson’s disease patients.

The analysis used the company’s reverse-engineering algorithms, as well as some 5,000 simulations to identify genetic factors and biomarkers in cerebrospinal fluid, particularly in older males, that predict faster progression of Parkinson’s disease. The team validated its initial findings with an older clinical study, drawing data on 317 of its participants. Those results also show the identified factors with statistically reliable predictive power for Parkinson’s disease progression.

In addition, the findings suggest the ability to better predict the progression of Parkinson’s disease can help refine the design and recruitment of clinical trials testing new therapies. The researchers say applying results of these models can reduce variability in the results and thus reduce the size of trial samples by as much as 20 percent.

“Being able to use these predictors in the clinical setting,” says Colin Hill, GNS Healthcare CEO in a company statement, “will lead to faster and significantly cheaper clinical trials and accelerate the availability of new Parkinson’s disease drugs for patients in need.”

More from Science & Enterprise:

*     *     *

1 comment to Algorithms ID Factors in Parkinson’s Progression